A description will be given by taking inspection of a semiconductor wafer, for instance. A semiconductor device is fabricated by repeating steps of transcribing a pattern formed in a photo-mask onto a semiconductor wafer in lithography processes and etching processes. In the fabrication process of semiconductor devices, qualities of the lithography processes, the etching processes, and the like, generation of contaminants, and so forth have large influence upon the yield of the semiconductor devices. Accordingly, in order to detect earlier or in advance such abnormalities and generation of poor quality during fabrication processes as described above, various kinds of apparatuses for inspecting patterns on the semiconductor wafer in the course of fabrication process have been used.
As methods of inspecting defects existing in patterns on the semiconductor wafer, an optical type defect inspecting apparatus and an electron beam type inspecting apparatus have been put into practice. In the former light is irradiated on the semiconductor wafer and defects are detected using an optical image. In the latter an electron beam is irradiated on the semiconductor wafer, a secondary signal of generated secondary electrons, back scattering electrons, and the like, is detected, the signal is converted into an image, and a difference between the image and a comparative image is detected as a defect.                In the aforementioned electron beam type inspecting apparatus, inspection is conducted based on an image obtained by irradiating an electron beam. Accordingly, when noises interfere in the obtained image, a false inspection result is output. Therefore, techniques for reducing the image noise and enhancing the accuracy of inspection have been known (see PATENT LITERATURE 1, for example). However, a technique for studying causes of generation of a noise to be reduced has not been developed yet. Various causes of generating an image noise can be enumerated such as interferences of noises such as mechanical, electrical, and external disturbance and a fault of the apparatus. Therefore, identification of causes of an image noise is very difficult and it requires a long time.        